© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    Jacobian LensNEW
    Natural Language
    Autoencoders
    NEW    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsAPI Community BlogPrivacy & TermsContact
    1. Home
    3. Joseph Bloom · Open Source Sparse Autoencoders for all Residual Stream Layers of GPT2-Small
    5. GPT2-Small
    7. Residual Stream
    9. 4-RES-JB
    11. 1256
    Prev
    Next
    INDEX
    Explanations

    phrases related to bodily injury or violence directed towards someone's head

    oai_token-act-pair · gpt-3.5-turbo

    references to the head, both literally and metaphorically

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 4-res-jb
    Configuration
    jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.hook_resid_pre
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    24,576
    Data Type
    torch.float32
    Hook Point
    blocks.4.hook_resid_pre
    Architecture
    standard
    Context Size
    128
    Dataset
    Skylion007/openwebtext
    Hook Point Layer
    4
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
    PsyNetMessage
    -0.75
     Vict
    -0.72
     Suc
    -0.70
     Mub
    -0.69
     mell
    -0.64
     Sparkle
    -0.63
     Parenthood
    -0.63
    Constructed
    -0.63
     Typh
    -0.62
    éĹĺ
    -0.62
    POSITIVE LOGITS
    scar
    1.15
    butt
    1.15
    canon
    1.14
    ache
    1.06
    quarter
    1.06
    gear
    1.05
    dress
    1.05
    wind
    1.04
    locks
    0.99
    phones
    0.97
    Activations Density 0.026%

    No Known Activations